I have the following code to train. The validation records are based on a column indicating that they are part of validation or not.
valid_idx = []
all_idx = []
with open('/content/total_training.csv', newline='') as csvfile:
reader = csv.reader(csvfile, delimiter=',')
for row in reader:
if row[74] == 'true':
valid_idx.append(reader.line_num - 1)
all_idx.append(reader.line_num - 1)
coord_labels, semantic_labels = [], []
for i in range(18):
coord_labels += [f'x{i+1}', f'y{i+1}', f'conf{i+1}']
semantic_labels += [f'sem{i+1}']
dls = TabularDataLoaders.from_csv(
'/content/total_training.csv',
y_names='corrected_person_position_type_id',
cont_names = coord_labels,
cat_names = semantic_labels,
procs = [Categorify, Normalize],
valid_idx = valid_idx,
bs=4096
)
The CSV also contains an ID value which is important for me (first column of the CSV).
After plotting the confusion matrix:
I would like to know the ID’s of the rows that were predicted as lying_on_floor, but were in fact sitting. I played around with learner.show_results but that doesn’t seem to be it. Is there an easy way to accomplish this?